Abstract A Study of Interactive 3D Point Location in a Computer Simulated Virtual Environme

A Study of Interactive 3D Point Location in a Computer Simulated Virtual Environment James Boritz and Kellogg S. BoothDepartment of Computer ScienceUniversity of British ColumbiaVancouver, British Columbia, Canada boritz@cs.ubc.ca ksbooth@cs.ubc.caAbstractThe present study investigated the ability to interactively locate points in a three dimensional computer environment using a six degree of freedom input device. Four different visual feedback modes were tested: fixed viewpoint monoscopic perspective, fixed viewpoint stereoscopic perspective, head-tracked monoscopic perspective and head-tracked stereoscopic perspective. Targets were located at plus and minus 10cm along the X, Y or Z axes, from a fixed starting location. Data about the time to complete the task and positioning accuracy (error) are gathered for each trial. In addition, subjective feedback regarding the apparatus, visual mode and task difficulty was solicited from subjects. The results indicate that stereoscopic performance is superior to monoscopic performance and that asymmetries exist both across and within axes. Head tracking had no appreciable effect upon performance. Subjective feedback regarding performance is usually consistent with objective measures, but some inconsistencies are present. IntroductionThree dimensional computer graphics, and commercial six degree of freedom input devices are becoming more widespread. It is important to understand the quality of interaction as these tech-nologies make new interactive 3D applications possible. Most interaction with computers today takes place in a two dimensional environment. This is not surprising when one considers that the most popular computer applications are word processors and spreadsheets. However, even when the application is three dimen-sional in nature, such as computer modelling or volume visualisa-tion, interaction still often takes place in two dimensions [2] [4] [5] [12].The intent of our research is to better understand how people perform interactive tasks in a computer simulated three dimen-sional computer environment using an isotonic six degree of free-dom (6DOF) position control input device. Studies have been performed in this area, but the results are somewhat inconsistent. Additional studies are required to enable the development of a model of task performance in a 3D environment.We are interested in how different target positions in conjunction with different display modes (monoscopic vs. stereoscopic) and different head tracking modes (active and inactive) affect a user’s ability to perform a variety of simple tasks. The goal of these studies is to establish a framework within which designers of in-teractive 3D applications can determine what kind of user per-formance is possible for a given interactive task. Alternately, knowledge of what variables affect performance would allow designers to determine how to assemble a task in order to achieve a desired performance level. By focusing on relatively simple tasks that are components of more sophisticated interactive tasks, we hope to apply our findings to a broad range of applications. Takemura et al. [15] had subjects perform a volume location task (1cm and 2cm cubes) using a field sequential stereo display and a large number of target locations. They found no significant differ-ence between target positions. Ware and Balakrishnan [17] had subjects perform a volume location task (1cm cube) with targets to the left or behind the starting location of the cursor using a head-tracked field sequential stereo display and found perform-ance in Z to be 10% slower than performance in X.Massimino et al. [11] studied 3DOF tracking using an isometric rate controller with a monoscopic display and found the error rate in Z to be much higher than in X and Y. Zhai [18] and Zhai and Milgram [19] have performed several studies of 3D docking, tracking and target acquisition tasks using a variety of controllers and control types in conjunction with monoscopic and field se-quential stereoscopic displays. They find that the stereoscopic display is superior to the monoscopic display and that asymme-tries in the amount of error exist across the X, Y and Z axes. Many factors can influence the ability to perform an interactive task in a 3D environment. Wanger et al. [16] have studied factors that affect the ability to perceive spatial relationships in static 3D images. Zhai studied a variety of different 3D input devices [18]. Chung [3]studied the use of a head mounted display for a beam targeting task and found that subjects did not investigate beam directions that put a great deal of physical strain on their necks. Liang et al. [10] discuss the effect of sensing lag in the Polhemus Isotrak™ and present a filtering technique for dealing with this issue.Sollenberger and Milgram [14] studied the ability of subjects to identify the visual network to which a specified point belonged in monoscopic and stereoscopic modes with and without rotation of the network. They found that rotation and stereoscopic viewing improved performance both individually and in combination. Arthur et al. [1] studied a similar task using monoscopic and stereoscopic displays with and without head-coupling. They found that the head-coupled stereoscopic mode resulted in the shortest response time and lowest error rate.The present study is the first of a series of planned investigations into human ability to interact in 3D environment using a 6DOF position input device. These studies will focus on two simple tasks: point location and interactive path tracing. Point location requires a subject to move the tip of a three dimensional pointer to a fixed point (not a volume) in the three dimensionalenviron-